The most expensive guest you'll ever check in is the one arriving with the wrong picture in their head. They booked expecting a romantic ocean-view suite. What you have is a comfortable standard room with a partial view. You never said otherwise — but an AI assistant did, and now the disappointment is sitting at your front desk.
That scenario is the core argument of a recent piece in Fortune by Teresa Mackintosh, chief executive of Aven Hospitality. Her observation is uncomfortable and precise: AI systems are increasingly shaping what travellers expect before any direct contact with the brand — and they are getting the details wrong.
The promise you never made
The examples are mundane and damaging. A restaurant the AI recommends turns out to be closed for renovation. An itinerary presented as seamless doesn't actually connect. A room type is described with a confidence the inventory never supported. None of it came from the hotel. All of it lands on the hotel.
As Mackintosh puts it, guests "rarely blame the platform or recommendation engine that shaped their perception. They blame the hotel." The accountability flows to the brand at the end of the chain, regardless of where the distortion started.
"The more intermediaries between brands and consumers, the more opportunity there is for distortion." — Teresa Mackintosh, Fortune
Being recommended is not the same as being right
It helps to separate two jobs that get lumped together. Getting an AI engine to mention and recommend you at all is GEO — Generative Engine Optimisation: the narrative work of being one of the three properties the model surfaces while the rest stay invisible. Making sure the facts inside that recommendation are accurate is a different discipline: AEO, Answer Engine Optimisation.
AEO works on the structured facts a model can extract about your property and answer directly — room types, real availability, rates, amenities, policies, the content that maps cleanly to a question's literal terms. When a traveller asks "does it have an ocean-view suite under $400 with free parking?", the engine returns a short, confident answer with your name attached. AEO governs whether that answer is true.
This is why visibility without accuracy is not an asset — it's a liability with reach. An engine that confidently recommends your property on bad facts isn't marketing; it's a complaint generator. The guest arrives, the gap is exposed at the desk, and the review names you, not the model.
AEO is how you close the gap
If the engine can't find an authoritative version of your facts, it will assemble one from OTA listings, stale review snippets, and inference. That synthetic version becomes the promise. You inherit the gap. AEO is the discipline of making sure the engine answers from you. Three moves matter most:
- Establish a single source of truth. Consolidate room types, real availability, rates, and operational details into one authoritative place — not scattered across a CMS, a PDF, and three OTA extranets that disagree with each other.
- Make the facts machine-readable. Accuracy an engine can't parse doesn't count. Schema markup, structured FAQ and policy data, clean amenity tags — the facts have to be exposed in a form a model can extract and cite, not buried in marketing prose.
- Monitor what AI actually says. You cannot correct a distortion you never see. Audit how the major assistants answer real constraint queries about your property, and treat every confident error as an operational defect to dispute and fix.
This is the same logic that has always governed distribution: the fewer intermediaries standing between you and the guest, the less your facts bend in transit. The difference is that the most influential intermediary is now an answer engine replying in full sentences — and it sounds authoritative whether it's right or wrong.
Visibility was the goal of the last decade. Accuracy is the goal of this one. The properties that win in an AI-mediated market won't be the loudest in the model's answer — they'll be the ones whose answer is true.
Your move: stop asking whether AI mentions you, and start governing what it says you are.